Embedded real-time implementation of a computational efficient optical flow extraction method for intelligent robot control applications

The main role of an autonomous car is tracking a path on a determinate distance, while being able to notice road signs and to avoid collisions. Essential parts of these functions are the sensors, which identify the elements in the vehicle's environment. Path following can be done by different ways, from which we will underline the use of a new method, based on video processing and Optical Flow extraction. The aim is to build a real-time system suitable for implementation on resource-restricted platforms. Experimental results with the embedded, real-time implemented - on a FPGA supported Raspberry Pi platform - method are given in the paper, put to use in a line-following mobile robot application with intelligent control. We also prove the applicability of the new method in the take-off and landing stabilization of autonomous UAVs.